chenlin/Mujoco_WASM
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #2990 from google-deepmind:mjwarp_doc_batched_model_field_warning

Browse Source 
PiperOrigin-RevId: 853420667
Change-Id: I540b36611485054851e497c8f74302ac0fbedd36
This commit is contained in:
Copybara-Service
2026-01-07 14:46:36 -08:00
parent f703a0e023 8e39d50496
commit f4beb95212
1 changed files with 8 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
doc/mjwarp/index.rst
+8 -2
View File
@@ -330,8 +330,14 @@ To enable batched simulation with different model parameter values, many :class:
have a leading batch dimension. By default, the leading dimension is 1 (i.e., ``field.shape[0] == 1``) and the same
value(s) will be applied to all worlds. It is possible to override one of these fields with a ``wp.array`` that has a
leading dimension greater than one. This field will be indexed with a modulo operation of the world id and batch
dimension: ``field[worldid % field.shape[0]]``. Importantly, the field shape should be overridden prior to
:ref:`graph capture <mjwGC>` (i.e., ``wp.ScopedCapture``)
dimension: ``field[worldid % field.shape[0]]``.
.. admonition:: Graph capture
:class: warning
The field array should be overridden prior to
:ref:`graph capture <mjwGC>` (i.e., ``wp.ScopedCapture``)
since the update will not be applied to an existing graph.
.. code-block:: python